summaryrefslogtreecommitdiffstats
path: root/VNFs/DPPD-PROX/packet_utils.c
blob: e06529c47cccdc30d823e7629553f7eda9cee532 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
/*
// Copyright (c) 2010-2020 Intel Corporation
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//     http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
*/

#include <rte_lcore.h>
#include <rte_hash.h>
#include <rte_hash_crc.h>
#include "task_base.h"
#include "lconf.h"
#include "prefetch.h"
#include "log.h"
#include "defines.h"
#include "handle_master.h"
#include "prox_port_cfg.h"
#include "packet_utils.h"

static inline int find_ip(struct ether_hdr_arp *pkt, uint16_t len, uint32_t *ip_dst)
{
	prox_rte_vlan_hdr *vlan_hdr;
	prox_rte_ether_hdr *eth_hdr = (prox_rte_ether_hdr*)pkt;
	prox_rte_ipv4_hdr *ip;
	uint16_t ether_type = eth_hdr->ether_type;
	uint16_t l2_len = sizeof(prox_rte_ether_hdr);

	// Unstack VLAN tags
	while (((ether_type == ETYPE_8021ad) || (ether_type == ETYPE_VLAN)) && (l2_len + sizeof(prox_rte_vlan_hdr) < len)) {
		vlan_hdr = (prox_rte_vlan_hdr *)((uint8_t *)pkt + l2_len);
		l2_len +=4;
		ether_type = vlan_hdr->eth_proto;
	}

	switch (ether_type) {
	case ETYPE_MPLSU:
	case ETYPE_MPLSM:
		// In case of MPLS, next hop MAC is based on MPLS, not destination IP
		l2_len = 0;
		break;
	case ETYPE_IPv4:
		break;
	case ETYPE_EoGRE:
	case ETYPE_ARP:
	case ETYPE_IPv6:
		l2_len = 0;
		break;
	default:
		l2_len = 0;
		plog_warn("Unsupported packet type %x - CRC might be wrong\n", ether_type);
		break;
	}

	if (l2_len && (l2_len + sizeof(prox_rte_ipv4_hdr) <= len)) {
		prox_rte_ipv4_hdr *ip = (prox_rte_ipv4_hdr *)((uint8_t *)pkt + l2_len);
		// TODO: implement LPM => replace ip_dst by next hop IP DST
		*ip_dst = ip->dst_addr;
		return 0;
	}
	return -1;
}

/* This implementation could be improved: instead of checking each time we send a packet whether we need also
   to send an ARP, we should only check whether the MAC is valid.
   We should check arp_update_time in the master process. This would also require the generating task to clear its arp ring
   to avoid sending many ARP while starting after a long stop.
   We could also check for arp_timeout in the master so that dataplane has only to check whether MAC is available
   but this would require either thread safety, or the the exchange of information between master and generating core.
*/

int write_dst_mac(struct task_base *tbase, struct rte_mbuf *mbuf, uint32_t *ip_dst, uint64_t **time)
{
	const uint64_t hz = rte_get_tsc_hz();
	struct ether_hdr_arp *packet = rte_pktmbuf_mtod(mbuf, struct ether_hdr_arp *);
	prox_rte_ether_addr *mac = &packet->ether_hdr.d_addr;

	uint64_t tsc = rte_rdtsc();
	struct l3_base *l3 = &(tbase->l3);
	if (l3->gw.ip) {
		if (likely((l3->flags & FLAG_DST_MAC_KNOWN) && (tsc < l3->gw.arp_update_time) && (tsc < l3->gw.arp_timeout))) {
			memcpy(mac, &l3->gw.mac, sizeof(prox_rte_ether_addr));
			return SEND_MBUF;
		} else if (tsc > l3->gw.arp_update_time) {
			// long time since we have successfully sent an arp, send arp
			// If sending ARP failed (ring full) then arp_update_time is not updated to avoid having to wait 1 sec to send ARP REQ again
			*time = &l3->gw.arp_update_time;
			*ip_dst = l3->gw.ip;
			if ((l3->flags & FLAG_DST_MAC_KNOWN) && (tsc < l3->gw.arp_timeout)){
				// MAC is valid in the table => send also the mbuf
				memcpy(mac, &l3->gw.mac, sizeof(prox_rte_ether_addr));
				return SEND_MBUF_AND_ARP;
			} else {
				// MAC still unknown, or timed out => only send ARP
				return SEND_ARP;
			}
		} else {
			// MAC is unknown and we already sent an ARP recently, drop mbuf and wait for ARP reply
			return DROP_MBUF;
		}
	}

	uint16_t len = rte_pktmbuf_pkt_len(mbuf);
	if (find_ip(packet, len, ip_dst) != 0) {
		// Unable to find IP address => non IP packet => send it as it
		return SEND_MBUF;
	}
	if (likely(l3->n_pkts < 4)) {
		for (unsigned int idx = 0; idx < l3->n_pkts; idx++) {
			if (*ip_dst == l3->optimized_arp_table[idx].ip) {
				 // IP address already in table
				if ((tsc < l3->optimized_arp_table[idx].arp_update_time) && (tsc < l3->optimized_arp_table[idx].arp_timeout)) {
					// MAC address was recently updated in table, use it
					memcpy(mac, &l3->optimized_arp_table[idx].mac, sizeof(prox_rte_ether_addr));
					return SEND_MBUF;
				} else if (tsc > l3->optimized_arp_table[idx].arp_update_time) {
					// ARP not sent since a long time, send ARP
						*time = &l3->optimized_arp_table[idx].arp_update_time;
					if (tsc < l3->optimized_arp_table[idx].arp_timeout) {
						// MAC still valid => also send mbuf
						memcpy(mac, &l3->optimized_arp_table[idx].mac, sizeof(prox_rte_ether_addr));
						return SEND_MBUF_AND_ARP;
					} else {
						// MAC unvalid => only send ARP
						return SEND_ARP;
					}
				} else {
					//  ARP timeout elapsed, MAC not valid anymore but waiting for ARP reply
					return DROP_MBUF;
				}
			}
		}
		// IP address not found in table
		l3->optimized_arp_table[l3->n_pkts].ip = *ip_dst;
		*time = &l3->optimized_arp_table[l3->n_pkts].arp_update_time;
		l3->n_pkts++;

		if (l3->n_pkts < 4) {
			return SEND_ARP;
		}

		// We have too many IP addresses to search linearly; lets use hash table instead => copy all entries in hash table
		for (uint32_t idx = 0; idx < l3->n_pkts; idx++) {
			uint32_t ip = l3->optimized_arp_table[idx].ip;
			int ret = rte_hash_add_key(l3->ip_hash, (const void *)&ip);
			if (ret < 0) {
				// This should not happen as few entries so far.
				// If it happens, we still send the ARP as easier:
				//      If the ARP corresponds to this error, the ARP reply will be ignored
				//      If ARP does not correspond to this error/ip, then ARP reply will be handled.
				plogx_err("Unable add ip %d.%d.%d.%d in mac_hash (already %d entries)\n", IP4(ip), idx);
			} else {
				memcpy(&l3->arp_table[ret], &l3->optimized_arp_table[idx], sizeof(struct arp_table));
			}
		}
		return SEND_ARP;
	} else {
		// Find IP in lookup table. Send ARP if not found
		int ret = rte_hash_lookup(l3->ip_hash, (const void *)ip_dst);
		if (unlikely(ret < 0)) {
			// IP not found, try to send an ARP
			int ret = rte_hash_add_key(l3->ip_hash, (const void *)ip_dst);
			if (ret < 0) {
				// No reason to send ARP, as reply would be anyhow ignored
				plogx_err("Unable to add ip %d.%d.%d.%d in mac_hash\n", IP4(*ip_dst));
				return DROP_MBUF;
			} else {
				l3->arp_table[ret].ip = *ip_dst;
				*time = &l3->arp_table[ret].arp_update_time;
			}
			return SEND_ARP;
		} else {
			// IP has been found
			if (likely((tsc < l3->arp_table[ret].arp_update_time) && (tsc < l3->arp_table[ret].arp_timeout))) {
				// MAC still valid and ARP sent recently
				memcpy(mac, &l3->arp_table[ret].mac, sizeof(prox_rte_ether_addr));
				return SEND_MBUF;
			} else if (tsc > l3->arp_table[ret].arp_update_time) {
				// ARP not sent since a long time, send ARP
				*time = &l3->arp_table[ret].arp_update_time;
				if (tsc < l3->arp_table[ret].arp_timeout) {
					// MAC still valid => send also MBUF
					memcpy(mac, &l3->arp_table[ret].mac, sizeof(prox_rte_ether_addr));
					return SEND_MBUF_AND_ARP;
				} else {
					return SEND_ARP;
				}
			} else {
				return DROP_MBUF;
			}
		}
	}
	// Should not happen
	return DROP_MBUF;
}

void task_init_l3(struct task_base *tbase, struct task_args *targ)
{
	static char hash_name[30];
	uint32_t n_entries = MAX_ARP_ENTRIES * 4;
	const int socket_id = rte_lcore_to_socket_id(targ->lconf->id);
	sprintf(hash_name, "A%03d_%03d_mac_table", targ->lconf->id, targ->id);

	hash_name[0]++;

	struct rte_hash_parameters hash_params = {
		.name = hash_name,
		.entries = n_entries,
		.key_len = sizeof(uint32_t),
		.hash_func = rte_hash_crc,
		.hash_func_init_val = 0,
	};
	tbase->l3.ip_hash = rte_hash_create(&hash_params);
	PROX_PANIC(tbase->l3.ip_hash == NULL, "Failed to set up ip hash table\n");

	tbase->l3.arp_table = (struct arp_table *)prox_zmalloc(n_entries * sizeof(struct arp_table), socket_id);
	PROX_PANIC(tbase->l3.arp_table == NULL, "Failed to allocate memory for %u entries in arp table\n", n_entries);
	plog_info("\tarp table, with %d entries of size %ld\n", n_entries, sizeof(struct l3_base));

	targ->lconf->ctrl_func_p[targ->task] = handle_ctrl_plane_pkts;
	targ->lconf->ctrl_timeout = freq_to_tsc(targ->ctrl_freq);
	tbase->l3.gw.ip = rte_cpu_to_be_32(targ->gateway_ipv4);
	tbase->flags |= TASK_L3;
	tbase->l3.core_id = targ->lconf->id;
	tbase->l3.task_id = targ->id;
	tbase->l3.tmaster = targ->tmaster;
	tbase->l3.seed = (uint)rte_rdtsc();
	if (targ->arp_timeout != 0)
		tbase->l3.arp_timeout = targ->arp_timeout;
	else
		tbase->l3.arp_timeout = DEFAULT_ARP_TIMEOUT;
	if (targ->arp_update_time != 0)
		tbase->l3.arp_update_time = targ->arp_update_time;
	else
		tbase->l3.arp_update_time = DEFAULT_ARP_UPDATE_TIME;
}

void task_start_l3(struct task_base *tbase, struct task_args *targ)
{
	const int NB_ARP_MBUF = 1024;
	const int ARP_MBUF_SIZE = 2048;
	const int NB_CACHE_ARP_MBUF = 256;

	struct prox_port_cfg *port = find_reachable_port(targ);
        if (port && (tbase->l3.arp_pool == NULL)) {
		static char name[] = "arp0_pool";
                tbase->l3.reachable_port_id = port - prox_port_cfg;
		if (targ->local_ipv4) {
			tbase->local_ipv4 = rte_be_to_cpu_32(targ->local_ipv4);
			register_ip_to_ctrl_plane(tbase->l3.tmaster, tbase->local_ipv4, tbase->l3.reachable_port_id, targ->lconf->id, targ->id);
        	}
		master_init_vdev(tbase->l3.tmaster, tbase->l3.reachable_port_id, targ->lconf->id, targ->id);
		name[3]++;
		struct rte_mempool *ret = rte_mempool_create(name, NB_ARP_MBUF, ARP_MBUF_SIZE, NB_CACHE_ARP_MBUF,
			sizeof(struct rte_pktmbuf_pool_private), rte_pktmbuf_pool_init, NULL, rte_pktmbuf_init, 0,
			rte_socket_id(), 0);
		PROX_PANIC(ret == NULL, "Failed to allocate ARP memory pool on socket %u with %u elements\n",
			rte_socket_id(), NB_ARP_MBUF);
		plog_info("\t\tMempool %p (%s) size = %u * %u cache %u, socket %d\n", ret, name, NB_ARP_MBUF,
			ARP_MBUF_SIZE, NB_CACHE_ARP_MBUF, rte_socket_id());
		tbase->l3.arp_pool = ret;
	}
}

void task_set_gateway_ip(struct task_base *tbase, uint32_t ip)
{
	tbase->l3.gw.ip = ip;
	tbase->flags &= ~FLAG_DST_MAC_KNOWN;
}

void task_set_local_ip(struct task_base *tbase, uint32_t ip)
{
	tbase->local_ipv4 = ip;
}

static void reset_arp_update_time(struct l3_base *l3, uint32_t ip)
{
	uint32_t idx;
	plogx_info("\tMAC entry for IP "IPv4_BYTES_FMT" timeout in kernel\n", IP4(ip));
	if (ip == l3->gw.ip) {
		l3->gw.arp_update_time = 0;
	} else if (l3->n_pkts < 4) {
		for (idx = 0; idx < l3->n_pkts; idx++) {
			uint32_t ip_dst = l3->optimized_arp_table[idx].ip;
			if (ip_dst == ip)
				break;
		}
		if (idx < l3->n_pkts) {
			l3->optimized_arp_table[idx].arp_update_time = 0;
		}
	} else {
		int ret = rte_hash_lookup(l3->ip_hash, (const void *)&ip);
		if (ret >= 0)
			l3->arp_table[ret].arp_update_time = 0;
	}
	return;
}

void handle_ctrl_plane_pkts(struct task_base *tbase, struct rte_mbuf **mbufs, uint16_t n_pkts)
{
	uint8_t out[1];
	const uint64_t hz = rte_get_tsc_hz();
	uint32_t ip, ip_dst, idx;
	int j;
	uint16_t command;
	struct ether_hdr_arp *hdr;
	struct l3_base *l3 = &tbase->l3;
	uint64_t tsc= rte_rdtsc();
	uint64_t update_time = l3->arp_timeout * hz / 1000;

	for (j = 0; j < n_pkts; ++j) {
		PREFETCH0(mbufs[j]);
	}
	for (j = 0; j < n_pkts; ++j) {
		PREFETCH0(rte_pktmbuf_mtod(mbufs[j], void *));
	}

	for (j = 0; j < n_pkts; ++j) {
		out[0] = OUT_HANDLED;
		command = mbufs[j]->udata64 & 0xFFFF;
		plogx_dbg("\tReceived %s mbuf %p\n", actions_string[command], mbufs[j]);
		switch(command) {
		case UPDATE_FROM_CTRL:
			hdr = rte_pktmbuf_mtod(mbufs[j], struct ether_hdr_arp *);
			ip = (mbufs[j]->udata64 >> 32) & 0xFFFFFFFF;

			if (prox_rte_is_zero_ether_addr(&hdr->arp.data.sha)) {
				// MAC timeout or deleted from kernel table => reset update_time
				// This will cause us to send new ARP request
				// However, as arp_timeout not touched, we should continue sending our regular IP packets
				reset_arp_update_time(l3, ip);
				plogx_info("\tTimeout for MAC entry for IP "IPv4_BYTES_FMT"\n", IP4(ip));
				return;
			} else
				plogx_dbg("\tUpdating MAC entry for IP "IPv4_BYTES_FMT" with MAC "MAC_BYTES_FMT"\n",
					IP4(ip), MAC_BYTES(hdr->arp.data.sha.addr_bytes));
			if (ip == l3->gw.ip) {
				// MAC address of the gateway
				memcpy(&l3->gw.mac, &hdr->arp.data.sha, 6);
				l3->flags |= FLAG_DST_MAC_KNOWN;
				l3->gw.arp_timeout = tsc + update_time;
				update_arp_update_time(l3, &l3->gw.arp_update_time, l3->arp_update_time);
			} else if (l3->n_pkts < 4) {
				// Few packets tracked - should be faster to loop through them thean using a hash table
				for (idx = 0; idx < l3->n_pkts; idx++) {
					ip_dst = l3->optimized_arp_table[idx].ip;
					if (ip_dst == ip)
						break;
				}
				if (idx < l3->n_pkts) {
					memcpy(&l3->optimized_arp_table[idx].mac, &(hdr->arp.data.sha), sizeof(prox_rte_ether_addr));
					l3->optimized_arp_table[idx].arp_timeout = tsc + update_time;
					update_arp_update_time(l3, &l3->optimized_arp_table[idx].arp_update_time, l3->arp_update_time);
				}
			} else {
				int ret = rte_hash_add_key(l3->ip_hash, (const void *)&ip);
				if (ret < 0) {
					plogx_info("Unable add ip %d.%d.%d.%d in mac_hash\n", IP4(ip));
				} else {
					memcpy(&l3->arp_table[ret].mac, &(hdr->arp.data.sha), sizeof(prox_rte_ether_addr));
					l3->arp_table[ret].arp_timeout = tsc + update_time;
					update_arp_update_time(l3, &l3->arp_table[ret].arp_update_time, l3->arp_update_time);
				}
			}
			tx_drop(mbufs[j]);
			break;
		case ARP_REPLY_FROM_CTRL:
		case ICMP_FROM_CTRL:
		case ARP_REQ_FROM_CTRL:
		case PKT_FROM_TAP:
			out[0] = 0;
			// tx_ctrlplane_pkt does not drop packets
			plogx_dbg("\tForwarding (ARP/PING) packet from master\n");
			tbase->aux->tx_ctrlplane_pkt(tbase, &mbufs[j], 1, out);
			TASK_STATS_ADD_TX_NON_DP(&tbase->aux->stats, 1);
			break;
		}
	}
}